Artificial Intelligence and the Impact of Computing: Unit Exam B
What AI is and is not, how a spam filter learns, unfair proxies such as postcodes, kinds of AI task, reducing bias, privacy and environmental costs.
Instructions
- Time allowed
- 25 minutes
- Total marks
- 25
- Calculator
- No calculator is permitted.
- Answer all the questions.
- The number of marks for each question is shown in brackets. Use it as a guide to how much to write.
- The clock at the top counts down from the real time allowed. When it reaches zero the paper is locked and marked, as it would be collected in the exam hall.
- Read every question to the end before answering. Where a question says tick two boxes, tick exactly two: ticking more than asked scores nothing. Type one word and number answers carefully, because they are marked exactly.
Questions
Answer every question, then finish the paper to have it marked. Marking as you go teaches you far less.
Tick one box.
- LC Computer systems that do tasks which normally need human intelligence, such as recognising speech
AI means tasks we would call intelligent if a person did them.
| True | False | |
|---|---|---|
| AI systems understand the meaning of what they do in the way people do | ||
| AI learns from examples in its training data | ||
| An AI can be biased even if nobody meant it to be | ||
| Using AI always reduces energy use |
Tick one box in each row.
- LAI systems understand the meaning of what they do in the way people do: False
- LAI learns from examples in its training data: True
- LAn AI can be biased even if nobody meant it to be: True
- LUsing AI always reduces energy use: False
AI finds patterns without understanding them, learns from data, can pick up bias, and training it uses a lot of energy.
- LIt is given many emails labelled as spam or not spam
- LIt learns the patterns, such as words, senders or links, that are common in spam
- LIt uses those patterns to sort new emails, and improves when users mark mistakes
The filter is trained on a large set of emails, each labelled spam or not spam. It learns which features, such as certain words, unknown senders and suspicious links, appear far more often in spam, and uses them to sort new emails. Each time a user marks an email as spam it gets more examples and improves.
Tick one box.
- LB The postcode can stand in for things such as income or ethnicity, so the AI copies existing inequality
A harmless looking feature can stand in for a protected one.
- LIt can automate routine jobs, such as checkout or call centre work, so those jobs are lost
- LIt also creates new jobs and means workers need retraining
AI can automate routine work such as answering customer calls or checking forms, so fewer people are needed for those jobs. At the same time it creates new jobs building and checking AI systems, so workers need retraining to move into them.
| Recognising images | Understanding speech | Making recommendations | Translating languages | |
|---|---|---|---|---|
| A voice assistant setting a timer when asked | ||||
| A shopping site suggesting products | ||||
| A camera app identifying a dog in a photo | ||||
| An app turning a French menu into English |
Tick one box in each row.
- LA voice assistant setting a timer when asked: Understanding speech
- LA shopping site suggesting products: Making recommendations
- LA camera app identifying a dog in a photo: Recognising images
- LAn app turning a French menu into English: Translating languages
Each is a different AI task: speech, recommendation, image recognition and translation.
Tick one box.
- LA Train it on varied, representative data and test it on different groups
Representative data and testing across groups catch bias before it does harm.
- LThey are always listening and may record conversations
- LThe data is stored by the company and may be shared, sold or used for adverts
- LThe devices could be hacked, letting strangers listen in
- LWeighs this against the convenience, and reaches a judgement such as using privacy settings
A smart speaker is always listening for its wake word, and recordings of what people say at home can be stored on the company's servers, shared with partners or used to target adverts. A device on the home network can also be hacked. They are genuinely convenient, so the sensible conclusion is not that nobody should own one, but that owners should use the mute button, delete recordings and check the privacy settings.
Tick one box.
- LD Using machines or software to do tasks without people
Automation replaces human effort with machines or software.
Tick two boxes.
- LA E-waste from devices that are thrown away
- LC Electricity used by data centres
- LNo marks if more than 2 boxes are ticked
Energy use and e-waste are costs; less paper and less travel are benefits.
- Ltraining data
Training data.
Tick one box.
- LD It learned from patterns in huge amounts of data, not from rules a person wrote
There is no list of rules to read: the behaviour is spread across what it learned.
This is an original Key Stage 3 assessment. Most questions are closed, so the marking is exact, and the written answers are marked against a mark scheme in the same style as a GCSE paper, so the format is familiar long before it counts.